obonet

• What it is — obonet is a lightweight Python package for reading OBO-serialized ontologies into a NetworkX MultiDiGraph. Its parser targets OBO format specifications 1.2 and 1.4, and the project describes itself as specializing in OBO-to-NetworkX loading rather than general ontology processing.

• Key concepts — Ontology terms become graph nodes, while typed relationships become directed edges; multiple relationships between the same nodes are retained by the MultiDiGraph. Traditional ontology edges such as is_a run from subterm to superterm, so in this representation NetworkX descendants returns superterms and ancestors returns subterms. The tutorial demonstrates node properties, ID/name mappings, parent and child relationships, all paths to a root, ontology-level metadata, and obsolete-term replacement via replaced_by.

• How you’d use it — Call obonet.read_obo() with a local path, URL, or open file handle; compression is inferred from the path extension. Analyze the returned graph with NetworkX—for example, count nodes and edges, check whether it is a DAG, traverse superterms/subterms, inspect properties, or enumerate paths to a root. Pass include_clauses=True to preserve parsed OBO comments and trailing modifiers, pass ignore_obsolete=False when building obsolete-to-replacement mappings, or use the CLI to convert OBO to NetworkX node-link JSON.

• LLM angle — none stated

• Pitfalls & lessons — Only reading OBO files is supported. Some ontology nodes may have an ID but no name, so mappings should access name defensively. The documented edge direction can be confusing because it makes NetworkX ancestor/descendant terminology opposite the ontology’s general/specific hierarchy. Compared with the more general nxontology/pronto route, format coverage and retained metadata differ; conversion to a plain DiGraph requires choosing relationship types, reversing edges, and collapsing parallel edges.

• Verdict — A focused choice when the job is to load OBO 1.2/1.4 data into a metadata-rich NetworkX multigraph and analyze it with standard graph operations.

Sources consulted

• README.md
• examples/go-obonet.ipynb